The Journal of Neuroscience, October 26, 2005 • 25(43):9825–9835 • 9825

Cellular/Molecular Regulation of Kv7 (KCNQ) Kϩ Channel Open Probability by 4,5-Bisphosphate

Yang Li,1 Nikita Gamper,1 Donald W. Hilgemann,2 and Mark S. Shapiro1 1Department of Physiology, University of Texas Health Science Center at San Antonio, San Antonio, Texas 78229, and 2Department of Physiology, University of Texas Southwest Medical Center, Dallas, Texas 75235-9040

Voltage-gated Kv7 (KCNQ) channels underlie important K ϩ currents, including the neuronal M current, and are thought to be sensitive to membrane phosphatidylinositol 4,5-bisphosphate (PIP2 ) and PIP2 depletion to underlie muscarinic inhibition. We studied regulation of Kv7.2–7.4 channels by PIP2 in Chinese hamster ovary (CHO) cells using single-channel and whole-cell patch clamp and biochemical analysis. Maximal open probabilities (Po ) of Kv7.2-Kv7.4 homomultimers and of Kv7.2/7.3 heteromultimers were found to be strongly dependent on the [diC8-PIP2] applied to inside-out patches, with differential apparent affinities that correlate with their maximal Po in on-cell mode. Unitary conductance was not affected by PIP2. Raising tonic [PIP2] by coexpression of phosphatidylinositol (4)5- increased the maximal Po of both Kv7.2 and Kv7.2/7.3 channels studied in on-cell patches and increased whole-cell Kv7.2, but not Kv7.3, current amplitudes. In cells coexpressed with muscarinic M1 receptors, bath application of muscarinic reduced the maximal Po of Kv7.2/7.3 channels isolated in on-cell patches. Coexpression of a PIP2 sequestering construct moderately reduced whole- cell Kv7.2/7.3 currents, and coexpression of a construct containing a PIP2 nearly abolished them. Finally, biochemical analysis of anionic in CHO cells stably expressing M1 receptors shows that PIP2 and PIP are nearly depleted 1 min after muscarinic stimulation, with an unexpected rebound after 10 min. These results strongly support the direct regulation of Kv7 channels by

PIP2 anditsdepletionasthemechanismofmuscarinicsuppressionofMchannels.DivergentapparentaffinitiesofKv7.2–7.4channelsfor PIP2 may underlie their highly differential maximal Po observed in cell-attached patches. Key words: signaling; potassium channel; M current; phosphatidylinositol 4,5-bisphosphate; single channel; gating; patch clamp

Introduction family of M-type voltage-gated K ϩ channels, made by Kv7.1–7.5 Many ion channels and transporters have been shown to be sen- (KCNQ1–5) subunits (Delmas and Brown, 2005). Most M cur- sitive to the presence of phosphatidylinositol 4,5-bisphosphate rents in are made by homomultimers and heteromul-

(PIP2) in the membrane (Suh and Hille, 2005). Particularly well timers of Kv7.2, Kv7.3, and Kv7.5 (Wang et al., 1998; Lerche et al., studied has been PIP regulation of the family of inward rectifier 2000; Cooper et al., 2001; Roche et al., 2002; Hadley et al., 2003); 2 ϩ K ϩ (Kir) channels, for which differential sensitivity to membrane Kv7.4 makes important K currents in the inner ear (Kubisch et

PIP2 has been suggested to underlie the highly divergent open al., 1999), and Kv7.1 co-assembles with auxiliary KCNE subunits probabilities of this channel family (Rohacs et al., 2003). In this to make the cardiac IKs current (Sanguinetti et al., 1996). Wide scenario, PIP2 has differential apparent affinities for each channel agreement is appearing that M channels are sensitive to the abun- , and a given channel with a high apparent affinity for PIP2 dance of membrane PIP2 and that activation of PLC, and subse- will display a high open probability (Po) at tonic PIP2 levels and quent PIP2 , can sufficiently deplete the membrane of be relatively insensitive to decreases in PIP2, whereas channels PIP2 to cause the observed downregulation of M currents by with a low apparent affinity display a much lower tonic Po and are stimulation of Gq/11-coupled M1 muscarinic acetylcholine recep- very sensitive to depletion of PIP2, as may happen after strong tors (mAChRs) (Suh and Hille, 2002; Ford et al., 2003; Zhang et activation of C (PLC). al., 2003; Suh et al., 2004). Although this hypothesis is compel- An emerging literature suggests that a similar permissive re- ling, single-channel demonstration of the direct gating of M quirement for membrane PIP controls the activity of the Kv7 2 channels by PIP2 has been lacking. We recently characterized the gating of Kv7.2–7.5 channels at the single-channel level and found them to have strikingly differ- Received June 23, 2005; revised Aug. 26, 2005; accepted Sept. 9, 2005. ent open probabilities in cell-attached patches (Li et al., 2004). This work was supported by National Institutes of Health Grants R01 NS043394 (M.S.S.) and R01 HL067942 (D.W.H.)andAmericanHeartAssociationPostdoctoralFellowship0325120Y(N.G.).WethankPamelaMartin,Kath- Whereas Kv7.3 has a maximal Po near unity, Kv7.2, Kv7.4, and ryn Boyd, and Chengcheng Shen for expert technical assistance and Brad Rothberg and James Stockand for helpful Kv7.5 have much lower maximal Po values and, in particular, discussions. Kv7.4 has a maximal Po over 10-fold lower than Kv7.3. Using a Correspondence should be addressed to Dr. Mark S. Shapiro, Department of Physiology, MS 7756, University of chimeric strategy, the highly divergent P values for Kv7.3 and Texas Health Science Center, 7703 Floyd Curl Drive, San Antonio, TX 78229. E-mail: [email protected]. o DOI:10.1523/JNEUROSCI.2597-05.2005 Kv7.4 were found to be conferred by the cytoplasmic C terminal Copyright © 2005 Society for Neuroscience 0270-6474/05/259825-11$15.00/0 of the channel. Here, we sought to more directly test the putative • ϩ 9826 • J. Neurosci., October 26, 2005 • 25(43):9825–9835 Li et al. PIP2 Regulation of Unitary KCNQ K Channels

PIP2 regulation of Kv7 channel gating at the single-channel and tential, the single-channel amplitude (i) was calculated by fitting all- whole-cell level and by biochemical analysis. We find that Kv7 point histograms with single or multi-Gaussian curves. The difference between the fitted “closed” and “open” peaks was taken as i. When su- channels are indeed highly sensitive to the PIP2 concentration at the cytoplasmic face of the patch and that the different Kv7 chan- perimposed openings were observed, the number of channels in the patch was estimated from the maximal number of superimposed open- nels display a highly differential apparent affinity for PIP2 that ings. The apparent NPo was estimated as follows: may underlie their differential Po in intact cells. Furthermore, changes in [PIP2] have the expected effect on channel activity N consistent with the PIP2-regulation hypothesis. Finally, PIP2 and NP ϭ ͩ͸ t jͪ T (1) o j Ͳ PIP levels are shown to dramatically decrease and then rebound jϭ1 partially in parallel with a continuing fall of phosphatidylinositol (PI) and rise of (PA). where tj is the time spent at each current level corresponding to j ϭ 0, 1, 2...N,T is the duration of the recording, and N is the number of current Materials and Methods levels (minimum number of active channels). cDNA constructs. Plasmids encoding human Kv7.2, rat Kv7.3, human Dose–response curves of channel Po versus [diC8-PIP2] were fit using Kv7.4, and human Kv7.5 (GenBank accession numbers AF110020, IgorPro software (WaveMetrics, Lake Oswego, OR) by the following Hill AF091247, AF105202, and AF249278, respectively) were kindly given to equation: us by D. McKinnon (State University of New York, Stony Brook, NY; n x Kv7.2 and Kv7.3), T. Jentsch (Zentrum fu¨r Molekulare Neurobiologie, ϭ Ͳͫ ϩ ͩ halfͪ ͬ Po Po,max 1 (2) Hamburg, Germany; Kv7.4), and K. Steinmeyer (Aventis Pharma, x Frankfurt am Main, Germany; Kv7.5). Kv7.2 and Kv7.3 were subcloned where P is set to 1, x is [diC8-PIP ], x is the [diC8-PIP ] at which into pcDNA3 (Invitrogen, San Diego, CA) as described previously (Sha- o,max 2 half 2 P ϭ 0.5, and n is the Hill coefficient. piro et al., 2000). Kv7.4 and Kv7.5 were subcloned into pcDNA3.1zeoϩ o Whole-cell electrophysiology. Whole-cell and perforated-patch experi- and pcDNA3.1zeoϪ (Invitrogen) using XhoI/HindIII and XbaI/EcoRI, ments were performed as described previously (Gamper et al., 2003). To respectively. EGFP-PLC␦-PH, EGFP-Lyn-PH-PP, and EGFP-Akt-PH evaluate the amplitude of macroscopic Kv7.2/7.3 currents, CHO cells constructs were kind gifts from T. Meyer (Stanford University, Stanford, were held at 0 mV, and 800 ms hyperpolarizing steps to Ϫ60 mV, fol- CA). Type 1␣ PI(4)P5-kinase [PI(4)5-kinase] was kindly given to us by L. lowed by 1 s pulses back to 0 mV, were applied. The amplitude of the Pott (Ruhr-University, Bochum, Germany). The Chinese hamster ovary current was usually defined as the difference between the holding current (CHO) cell line stably expressing M muscarinic receptors was kindly 1 at 0 mV and the current at the beginning (after any capacity current as given to us by Prof. D. A. Brown (University College London, London, subsided) of the 1 s pulse back to 0 mV. In some cells, a more precise UK). The human mAChR M clone was purchased from the Guthrie 1 measurement was the XE991- or linopirdine-sensitive current at the Foundation (Sayre, PA). holding potential of 0 mV. CHO cells have negligible endogenous mac- Cell culture and transfections. CHO cells were used for electrophysio- ϩ roscopic K currents under our experimental conditions, and 50 ␮M logical analysis as described recently (Gamper et al., 2005a). Cells were ϩ XE991 or linopirdine completely blocked the K current in Kv7- grown in 100 mm tissue culture dishes (Falcon; Becton Dickinson, transfected CHO cells but had no effect on currents in nontransfected Mountain View, CA) in DMEM with 10% heat-inactivated fetal bovine cells (Gamper et al., 2005a). All results are reported as mean Ϯ SEM. and 0.1% penicillin and streptomycin in a humidified incubator at Phosphoinositide analysis. Anionic phospholipids were extracted from 37°C (5% CO ) and passaged every 3–4 d. Cells were discarded after ϳ30 2 cells, deacylated, and quantified from the glycerol head groups by sup- passages. For transfection, cells were plated onto poly-L-lysine-coated pressed conductivity as described previously (Nasuhoglu et al., 2002). coverslip chips and transfected 24 h later with Polyfect reagent (Qiagen, Hilden, Germany) according to the instructions of the manufacturer. For electrophysiological experiments, cells were used 48–96 h after transfec- Results tion. As a marker for successfully transfected cells, cDNA encoding green PIP2 stabilizes opening of unitary Kv7.2/7.3 channels and fluorescent protein (GFP) was cotransfected together with the cDNAs of reactivates “run-down” channels the of interest. We found that Ͼ95% of green-fluorescing cells We first examined the sensitivity of Kv7.2/7.3 (KCNQ2/3) het- expressed KCNQ currents in control experiments. eromultimers to PIP2, because these heteromeric channels com- Cell-attached/inside-out patch/single-channel electrophysiology. The prise most “M-type” channels in sympathetic ganglia cells (Wang methods of recording and analysis were similar to those previously used et al., 1998; Roche et al., 2002). Cloned Kv7.2ϩ7.3 subunits were for studying unitary Kv7 channels (Li et al., 2004). Channel activity was expressed in CHO cells and recordings from channels isolated in recorded 48–72 h after transfection in the cell-attached configuration. cell-attached patches, followed by excised inside-out patches. ⍀ Pipettes had resistances of 7–15 M when filled with a solution of the Heterologously expressed Kv7.2/7.3 channels produce voltage- following composition (in mM): 150 NaCl, 5 KCl, 1 MgCl , and 10 ϩ 2 gated K currents very similar to native M currents in neurons HEPES, pH 7.4 with NaOH. Cells were bath perfused with a solution of (Wang et al., 1998; Selyanko et al., 2000; Shapiro et al., 2000). the following composition (in mM): 175 KCl, 4 MgCl2, and 10 HEPES, pH 7.4 with KOH. There was no Ca 2ϩ nor Ca 2ϩ-chealator added to Because the muscarinic inhibition of M-type channels, thought 2ϩ these solutions, and we assume free Ca to be in the range of ϳ50 ␮M. to be attributable to depletion of PIP2 (Delmas and Brown, 2005), The resting membrane voltage was assumed to be 0 mV. Currents were is voltage independent (Shapiro et al., 2000), we focus here on the recorded using an Axopatch 1-D amplifier (Molecular Devices, Union Po of the channels at saturating voltages, which we refer to as their City, CA), sampled at 5 kHz, and filtered at 200–500 Hz (Kv7.4) or at maximal Po. This value was measured usually at 0 mV, which is a 0.5–1 kHz (all others). The short-chain, water-soluble PIP2 analog diC8- saturating voltage for these channels (Li et al., 2004). PIP (Echelon Biosciences, Salt Lake City, UT) was dissolved in an aque- 2 For channels sensitive to membrane PIP2 abundance, their ous stock solution at 10 mM and sonicated immediately before each use. activity in cell-attached patches should reflect the tonic level of Single-channel data were analyzed using PulseFit and TAC (Bruxton, PIP , channel run-down during excision the loss of PIP in the Seattle, WA) software. Open and closed events were analyzed by using 2 2 the “50% threshold criterion.” All events were carefully checked visually patch by membrane-associated , and channel activ- ity in inside-out patches should be governed by the PIP2 concen- before being accepted. Po histograms were generated using TACFit (Bruxton). The presence of only one channel in a patch was assumed if tration ([PIP2]) in the cytoplasmically facing bath solution. Be- Ͼ Ͼ the Po was 0.25 for 1 min without superimposed openings, especially cause full-length PIP2 itself is water insoluble, making at strongly depolarized potentials at which Po is highest. At a given po- quantification of its concentration in aqueous solution uncer- • ϩ Li et al. PIP2 Regulation of Unitary KCNQ K Channels J. Neurosci., October 26, 2005 • 25(43):9825–9835 • 9827

may underlie these differences. Figure 2 shows representative experiments at a patch potential of 0 mV, which is a saturat- ing voltage for these channels. A patch containing a single Kv7.2 channel (Fig. 2A) was excised and ran down to the mar- ␮ ginal Po value of 0.01. Application of 3 M diC8-PIP2 failed to significantly increase channel activity, but application of 100 ␮M

had a robust effect, increasing channel Po to 0.35. Records from a patch containing a unitary Kv7.3 channel are shown in Figure 2B. In on-cell mode, channel activity was near unity, as is usual for this channel (Li et al., 2004). After excision, channel activity somewhat ran down, but remained sub-

stantial, with a Po value of 0.24. Applica- ␮ tion of only 3 M diC8-PIP2 increased the ␮ Po to 0.72, and 30 M diC8-PIP2 sufficed to restore channel Po to near its value in on- cell mode. A patch containing a single Kv7.4 channel, on the other hand, exhib-

ited a negligible Po after excision to inside- out mode (0.001), and this value was only marginally increased by application of 20 ␮ ϭ M diC8-PIP2 (Po 0.005). However, the increased concentration in diC8-PIP2 to 150 ␮M resulted in a dramatic increase in channel P , to 0.43. Figure1. DiC8-PIP2 stabilizestheopeningofKv7.2/7.3channels.A,CurrentrecordsfromapatchcontainingasingleKv7.2/7.3 o channel in cell-attached mode (On-cell) and in inside-out mode with various concentrations of diC8-PIP2 in the cytoplasmically These types of experiments on homo- facing bath solution. B, Each 3 s sweep during the experiment was analyzed for channel Po, and the time course of the Po during meric Kv7.2–7.4 channels were performed the experiment is plotted. The transition from on-cell to inside-out mode is indicated by the arrow, and the applications of over a range of diC8-PIP2 concentrations, different diC8-PIP2 concentrations are indicated by the bars. and the data were summarized as a dose–

response relationship as Po versus diC8- tain, we used the short-chain, water-soluble analog dioctanoyl PIP2 concentration (Fig. 3A). For Kv7.2–7.4 homomultimers and (diC8)-PIP2, as have other investigators studying the PIP2 sensi- Kv7.2/7.3 heteromultimers, channel Po was well described by tivity of these channels (Zhang et al., 2003). Figure 1 shows a such a dose–response relationship with diC8-PIP2 concentration, representative experiment on a patch containing a single Kv7.2/ but with dramatically differential apparent affinities. For all of the

7.3 channel examined at a saturating voltage of 0 mV, which is channels, channel Po near unity could be achieved by application sufficiently positive to induce maximal Po (Li et al., 2004). In of diC8-PIP2, but at very different concentrations. For Kv7.2/7.3, Ϯ ␮ ϭ on-cell mode, the activity of the channel was present, but modest. Kv7.2, Kv7.3, and Kv7.4, the EC50 values were 40 6 M (n After excision, the channel ran down considerably, displaying 1–4 patches), 205 Ϯ 36 ␮M (n ϭ 1–8 patches), 2.6 Ϯ 0.5 ␮M (n ϭ only rare open events. The channel activity, however, was 1–7 patches), and 215 Ϯ 19 ␮M (n ϭ 1–3 patches), respectively.

strongly reactivated by the addition of diC8-PIP2 to the bath, Thus, there is a 100-fold difference in the apparent affinity of resulting in channel activity much greater than that exhibited in Kv7.3 versus Kv7.2 or Kv7.4 to diC8-PIP2, and we suppose also to on-cell mode. The modest channel activity in on-cell mode in this full-length PIP2, and heteromeric Kv7.2/7.3 channels have an patch probably reflects a low tonic PIP2 abundance in this partic- intermediate apparent affinity, as one would expect for a hetero- ␮ ular cell. At 50 M diC8-PIP2, the channel Po was almost 0.5, and meric channel composed of Kv7.2 and Kv7.3 subunits that both ␮ application of 200 M diC8-PIP2 caused the channel to remain contribute to PIP2 binding, and to gating. For the four types of open almost all of the time. For Kv7.2/7.3 channels isolated in channels, the Hill coefficients were 1.24 Ϯ 0.24, 1.33 Ϯ 0.15, Ϯ ϭ Ϯ Ϯ patches, the maximal Po in on-cell mode was 0.31 0.04 (n 0.97 0.17, and 1.74 0.15, respectively, very near that reported Ϯ 41); after excision, the channels ran down to a Po of 0.14 0.04 previously for diC8-PIP2 action on Kv7.2/7.3 (Zhang et al., 2003) ϭ ␮ (n 18), and the addition of 50 M diC8-PIP2 reactivated the and slightly lower than that modeled for full-length PIP2 action Ϯ ϭ channels to a Po of 0.65 0.09 (n 8). An additional increase in on Kv7.2/7.3 channels (Suh et al., 2004). ␮ diC8-PIP2 to 200 M increased the Po at 0 mV to near unity (0.95; The differential apparent affinities of diC8-PIP2 for activation ϭ n 2). of Kv channels parallels their maximal Po in cell-attached patches, for which the rank order is Kv7.3 Ͼ Kv7.2/7.3 Ͼ Kv7.2 Ͼ Kv7.4

Kv7.2–7.4 channels are sensitive to PIP2 with different (Li et al., 2004), suggesting that the differential Po of channels in apparent affinities on-cell mode may be governed by their differential apparent af-

We then similarly investigated the PIP2 sensitivity of Kv7.2–7.4 finity for plasma-membrane PIP2. By correlating the maximal Po homomultimers. We have previously showed Kv7.2–7.4 homo- of Kv7 channels in on-cell mode with their values as a function of

multimers to have a very different maximal Po (Li et al., 2004), diC8-PIP2 concentration, we can estimate what diC8-PIP2 con- raising the possibility that a differential apparent affinity for PIP2 centration correlates with tonic [PIP2] in intact cells. In cell- • ϩ 9828 • J. Neurosci., October 26, 2005 • 25(43):9825–9835 Li et al. PIP2 Regulation of Unitary KCNQ K Channels

attached patches, the maximal Po of Kv7.2/ 7.3, Kv7.2, Kv7.3, and Kv7.4 are 0.30–0.31, 0.04–0.17, 0.59–0.89, and 0.07 (Selyanko et al., 2001; Li et al., 2004; present study), respectively. From these values, we can estimate that an applied ϳ ␮ diC8-PIP2 concentration of 23 M in inside-out patches corresponds to the

tonic [PIP2] seen by channels in intact cells, because this concentration gives Po values similar to that seen in on-cell patches for all of the channels. We pre- sume that the affinity of the channels to

PIP2 differs from that for the shorter-chain diC8-PIP2 analog but that the relative ap- parent affinities for the different channels are likely similar for the two phosphoi- nositides. We also investigated whether

PIP2 changes the single-channel conduc- tance of Kv7 channels. Kv7.2/7.3 channels were recorded in inside-out patches over a range of potentials, both in the presence or ␮ absence of 30 M diC8-PIP2. The presence of the phosphoinositide, although having

a dramatic effect on channel Po, did not affect their unitary conductance, which was 6.1 Ϯ 0.4 pS (n ϭ 5) and 6.8 Ϯ 0.4 pS (n ϭ 5) in the absence or presence of 30 ␮M

diC8-PIP2, respectively (Fig. 3B). Figure2. A–C,IndividualsweepsfrompatchescontainingasingleKv7.2(A),Kv7.3(B),orKv7.4(C)channelinthepresenceof The molecule we call PIP2 is not the the indicated concentration of diC8-PIP2. The records were filtered at 500 Hz (A, B) and 200 Hz (C). only doubly phosphorylated phosphoino- sitide used as a membrane signaling molecule; the other is phos- thus reasoned that if Kv7 channels are indeed sensitive to mem-

phatidylinositol 3,4-bisphosphate [PI(3,4)P2]. The latter is pro- brane [PIP2], then its tonic increase by overexpression of PI(4)5- Ј duced by activation of phosphatidylinositol 3 -kinase (PI3- kinase should increase channel Po recorded in cell-attached kinase), typically as a result of receptor activation, patches. CHO cells were coexpressed with Kv7.2ϩKv7.3 or only and is well known for activation of the kinase Akt/PKB. Although Kv7.2, either alone or together with PI(4)5-kinase. For both

plasma-membrane PI(3,4)P2 concentrations are typically only a Kv7.2/7.3 and Kv7.2 channels, there was a dramatic increase in small fraction of [PIP2], even after strong PI3-kinase activation, maximal Po in cells coexpressed with PI(4)5-kinase. In addition, the sensitivity of other PIP2-sensitive channels to [PI(3,4)P2] has the profound run-down in channel activity observed after exci- been investigated as a probe of phosphoinositide binding (Ro- sion of patches from control cells was sharply reduced in patches

hacs et al., 1999). We tested whether bath application of PI(3,4)P2 from PI(4)5-kinase-transfected cells (Fig. 4A). These data are to inside-out patches containing Kv7.2/7.3 channels would also summarized in Figure 4B. In on-cell patches containing Kv7.2 ␮ Ϯ ϭ stabilize channel activity. We found that at 10 M, PI(3,4)P2 re- channels only, the Po at 0 mV was 0.035 0.013 (n 12), but in Ϯ activated channels to a similar maximal Po observed in cell- patches from cells coexpressing PI(4)5-kinase, the Po was 0.74 attached patches, and at 20 ␮M, channel activity was further in- 0.09 (n ϭ 10). In on-cell patches containing Kv7.2/7.3 channels Ϯ ϭ creased (data not shown). Thus, like certain types of Kir3 only, the Po at 0 mV was 0.31 0.04 (n 41), but in patches from Ϯ ϭ channels (Rohacs et al., 1999), Kv7.2/7.3 channels may have a cells coexpressing PI(4)5-kinase, the Po was 0.95 0.03 (n 13). nonspecific phosphoinositide activation profile, at least in terms These results are in accord with recent work from the Brown of these two distinct moieties of doubly phosphorylated laboratory, which showed that overexpression of PI(4)5-kinase

phosphoinositides. increased [PIP2] nearly threefold in sympathetic neurons, result- ing in strongly blunted muscarinic modulation of M currents

Increasing tonic [PIP2] by overexpression of PI(4)5-kinase (Winks et al., 2005). Although we cannot quantify the increase increases on-cell channel activity produced here, we estimate it to be at least that much, judging

Having ascertained that Kv7 channels are sensitive to plasma- from the robust increase in on-cell channel Po and from the membrane PIP2, we then manipulated [PIP2] in intact cells in strongly reduced run-down of the channels after patch excision. several different ways and examined the effects on maximal Po. Heterologously expressed homomeric Kv7.2 and 7.3 channels PIP2 is produced in the plasma membrane by sequential phos- display whole-cell current amplitudes some 10-fold smaller than phorylation of PI by PI4-kinase and PI(4)5-kinase. Overexpres- do Kv7.2/7.3 heteromultimers (Wang et al., 1998; Gamper et al., sion of PI(4)5-kinase in atrial myocytes or in sympathetic neu- 2003; Maljevic et al., 2003; Schwake et al., 2003). Although it has

rons has been shown to tonically increase membrane PIP2 and to been suggested that this difference is attributable to differential dramatically reduce desensitization of Kir3 channels in the surface expression of the homomeric and heteromeric channels former case and muscarinic modulation of endogenous M cur- (Schwake et al., 2000), a number of factors are likely involved

rent in the latter case (Bender et al., 2002; Winks et al., 2005). We (Etxeberria et al., 2004). The dramatic increase in maximal Po of • ϩ Li et al. PIP2 Regulation of Unitary KCNQ K Channels J. Neurosci., October 26, 2005 • 25(43):9825–9835 • 9829

Po of Kv7.3 homomultimers (Li et al., 2004) and its high apparent affinity for diC8-PIP2 (present study) predicts that overexpres- sion of PI(4)5-kinase should not increase macroscopic Kv7.3 cur- rents, and this was indeed the case as well (Fig. 4C). In control cells, Kv7.3 current densities were 7.0 Ϯ 1.5 pA/pF (n ϭ 10), and in cells coexpressed with PI(4)5-kinase, they were still low, at 4.9 Ϯ 1.0 pA/pF (n ϭ 10). These data reinforce the conclusion that Kv7.2 and Kv7.3 channels have highly differential apparent

affinities for PIP2 and furthermore suggest that part of the differ- ence between Kv7.2 and Kv7.2/7.3 current amplitudes may be

attributable to the lower apparent affinity of Kv7.2 for PIP2. However, the high affinity of Kv7.3 for diC8-PIP2 found in this study, reinforced by the lack of increase in whole-cell currents in cells overexpressing PI(4)5-kinase, suggests that the low macro- scopic expression Kv7.3 is not attributable to weak binding to

PIP2 and indicates a further difference between Kv7.2 and Kv7.3.

Cell-attached channel Po is suppressed by bath application of muscarinic

To further explore PIP2 regulation of Kv7 channels, we reduced membrane PIP2 abundance in intact CHO cells by receptor stim- ulation and by tonic PIP2 depletion. The separation of types of G-protein-mediated modulation of channels between “direct” and “second-messenger-mediated” mechanisms is best observed using cell-attached patch experiments. In direct mechanisms in- volving interactions between G-protein and channels, bath appli- cation of receptor agonists does not modulate the activity of channels isolated in the patch membrane, presumably because activated G-protein subunits cannot diffuse in the plane of the membrane past its interface with the glass of the pipette tip. How- ever, when second-messenger molecules underlie the modula-

tion, as is the case for Gq/11 actions, channels in the patch can be modulated by bath-applied agonists. For the case of muscarinic modulation of N-type Ca 2ϩ channels in SCG neurons, the rapid inhibition by G-protein ␤␥ dimers, thought to underlie modula-

tion by Go/i-coupled M2/4 receptors (Herlitze et al., 1996), is not seen in such cell-attached patch experiments, but the slower in-

hibition by Gq/11-coupled M1 receptors does affect channels in the patch, implicating an intracellular second messenger in the action (Hille, 1994). The latter pathway has been shown to be

attributable to depletion of PIP2 in the membrane (Gamper et al., 2004). Likewise, the M1 receptor-mediated modulation of M channels that gives M current its name also reaches past the glass/ membrane interface, suppressing M channels in the cell-attached

Figure 3. A, Dose–response relationship of the Po for Kv7.2 (squares), Kv7.3 (triangles), patch (Selyanko et al., 1992). Finally, M1 receptor-mediated de- Kv7.4 (inverted triangles), and Kv7.2/7.3 (circles) channels as a function of diC8-PIP2 concen- sensitization of GIRK (G-protein-activated inwardly rectifying tration in inside-out patches. The data were fit by a Hill equation (see Materials and Methods), K ϩ; Kir3) channels, thought also to be attributable to depletion with data errors taken into account by the fitting routine for all channels, except Kv7.2/7.3. The of membrane PIP , is also evident on channels in the patch when parameters of the fits are given in the text. B, The unitary current amplitudes for Kv7.2/7.3 2 ␮ M1 receptors in the rest of the cell are stimulated (Zhang et al., channels over a range of voltages, either in the absence or presence of diC8-PIP2 (30 M)inthe bath, are plotted. The data were fit linearly, yielding single-channel conductances given in the 2003). Thus, it seems likely that, unlike membrane , text. membrane PIP2 can diffuse into and out of the membrane in the tips of patch pipettes and is the exemplar case of a type signal recently termed “diffusible membrane delimited” (Delmas et al., Kv7.2 seen in cell-attached patches from PI(4)5-kinase- 2005). expressing cells suggested that part of the reason for the low mac- We wanted to analyze the muscarinic suppression of Kv7.2/ roscopic expression of Kv7.2 channels might be as a result of its 7.3 channels, mediated by depletion of PIP2, at the single-channel low apparent affinity for PIP2 and that such expression might be level, to compare it to regulation of the channels by diC8-PIP2 in much larger in cells with higher tonic [PIP2] produced by over- inside-out patches. CHO cells transfected with Kv7.2, Kv7.3, and expression of PI(4)5-kinase. Indeed, this turned out to be the case the human M1 receptor were studied in cell-attached patches, and (Fig. 4C). Control cells displayed very low Kv7.2 current densities unitary currents recorded before, during, and after application of (4.5 Ϯ 1.1 pA/pF; n ϭ 10), but those coexpressed with PI(4)5- the muscarinic agonist oxotremorine-M (oxo-M) (Fig. 5). Al- Ϯ kinase displayed current densities some eightfold larger (34.7 though such stimulation of heterologously expressed M1 recep- ϭ Ͻ 2ϩ 5.8 pA; n 10; p 0.001). On the other hand, the high maximal tors likely raises [Ca ]i (Shapiro et al., 2000), we have shown • ϩ 9830 • J. Neurosci., October 26, 2005 • 25(43):9825–9835 Li et al. PIP2 Regulation of Unitary KCNQ K Channels

Kv7.2/7.3 channels heterologously ex- pressed in CHO cells to be mostly insensi- 2ϩ tive to rises in [Ca ]i in the absence of coexpressed calmodulin (Gamper and Shapiro, 2003). In the experiment shown in Figure 5A, Po was 0.40 in control, slightly higher than that typically seen for Kv7.2/7.3 channels. Bath application of oxo-M (10 ␮M) strongly reduced channel

Po within several minutes to 0.07, and channel activity was reversible to a Po of 0.24 several minutes after washout of the agonist. In all such patches tested, the Po in control was 0.22 Ϯ 0.06; after application of oxo-M, the Po was sharply reduced by 73% to 0.060 Ϯ 0.018, and after washout, it mostly recovered, to 0.19 Ϯ 0.03 (n ϭ 4). This action of bath-applied muscarinic agonists on unitary Kv7.2/7.3 channels isolated in the cell-attached patch corre- sponds very closely with the action on whole-cell currents seen previously (Sely- anko et al., 2000; Shapiro et al., 2000). These data suggest that, although there are known differences in this signaling path- way between a heterologous expression system and real neurons, muscarinic mod- ulation of neuronal M currents can be wholly accounted for by a modulation of channel Po. It also reinforces the conclu- sion that PIP2 can diffuse into and out of patches of membrane in the tips of patch pipettes. From similarly transfected CHO cells, we also used “macropatches,” con- taining many Kv7.2/7.3 channels to ob- serve such inhibition from bath-applied agonist. These experiments also included the membrane-permeant IP3-receptor blocker 2-aminoethoxydiphenyl borate (Maruyama et al., 1997) to exclude the possibility that the inhibition was attribut- 2ϩ able to rises in [Ca ]i. Bath application of oxo-M (10 ␮M) strongly and reversibly in- hibited the Kv7.2/7.3 currents in the patches (data not shown), similar to previ- ous results using such macropatch experi- ments (Selyanko et al., 2000).

Coexpression of PIP2 sequestering or Figure 4. Overexpression of PI(4)5-kinase increases the Po of Kv7 channels in patches and increases whole-cell Kv7.2 but not degrading constructs reduces whole-cell Kv7.3,currents.A,IndividualsweepsfromapatchcontainingeitherasingleKv7.2/7.3oraKv7.2channelfromcontrolCHOcellsor Kv7.2/7.3 currents in CHO cells thosecotransfectedwithPI(4)5-kinaseareshown.Recordingswerefirstmadeinon-cellmodes,followedbyexcisiontoinside-out modes. O, Open channel levels; C, closed channel levels. B, Bars show the mean Po at 0 mV for all of the on-cell patches. C, Bars We also assayed the PIP2 sensitivity of Kv7.2/7.3 channels by testing several en- show whole-cell current amplitudes from CHO cells transfected with Kv7.2 or Kv7.3 alone or together with PI(4)5-kinase. Traces from representative experiments are displayed above. The voltage protocol used is as shown. hanced GFP (EGFP)-tagged constructs that have been shown to bind and seques- activated by, PI3-kinase-generated phosphoinositol PI(3,4)P ter PIP2 or to decrease its plasma membrane concentration. The 2 first, PLC␦-PH, contains the “pleckstrin homology” (PH) do- but not PI(4,5)P2 (the molecule in this work abbreviated as PIP2) ␦ ␦ main of PLC that binds to PIP2 to achieve its plasma-membrane (Franke et al., 1997), and so Akt-PH is a useful control for PLC - targeting (Shaw, 1996). Overexpression of PLC␦-PH has been PH. The third, Lyn-PH-PP, codes for a PIP2-specific phospho- Ј shown to sequester PIP2 and thus make it unavailable to act as an lipid, 5 -phosphatase (Stolz et al., 1998), that selectively reduces intracellular effector (Stauffer et al., 1998; Raucher et al., 2000; plasma membrane PIP2 concentrations but not other intracellu- Gamper et al., 2004). The second, Akt-PH, contains the PH do- lar pools of PIP2 or other plasma membrane PI phosphates, ex- main of the serine/threonine kinase Akt. Akt binds to, and is pressed as a fusion protein to a myristoylation/palmitoylation • ϩ Li et al. PIP2 Regulation of Unitary KCNQ K Channels J. Neurosci., October 26, 2005 • 25(43):9825–9835 • 9831

Figure6. ExpressionofPIP2 sequesteringordegradingconstructsreduceswhole-cellKv7.2/ 7.3 current amplitudes. CHO cells were cotransfected with Kv7.2ϩKv7.3 together with either GFPonly(control)orthePLC␦PH,Lyn-PH-PP,orAkt-PHconstructs.Currentswereelicitedusing theindicatedvoltageprotocolinwhole-cellconfiguration.A,Plottedaresuperimposedcurrent tracesfromfourdifferentcellstransfectedwitheachdifferentconstruct.B,Barsshowthemean whole-cell current amplitudes for these four groups of cells. **p Ͻ 0.01; ***p Ͻ 0.001.

their distribution within the cell (i.e., plasma membrane or cytoplasm). CHO cells were cotransfected with Kv7.2/7.3 channels and only GFP, Akt-PH, PLC␦-PH, or Lyn-PH-PP. For all four, we observed robust fluorescence that was strongly membrane local- ized for the cases of PLC␦-PH and Lyn-PH-PP or diffusely cyto- plasmic for the case of GFP and Akt-PH. We found that cotrans- fection of PLC␦-PH strongly reduced the amplitude of the Kv7.2/7.3 currents and Lyn-PH-PP nearly abolished the currents, Figure 5. Stimulation of M1 receptors by bath application of agonist reduces the Po of chan- ϩ but cotransfection of Akt-PH had little effect (Fig. 6). For these nels in cell-attached patches. CHO cells were cotransfected with Kv7.2 7.3 channels and M1 receptors. A, Sweeps show channel activities from a patch containing two Kv7.2/3 channels four groups of cells, the mean current densities were 58.3 Ϯ 9.4 ␮ ϭ Ϯ ϭ Ϯ Ͻ before, during, and after bath application of oxo-M (10 M). B, Bars show the mean unitary Po pA/pF (n 12), 69.9 14 pA/pF (n 9), 22.4 6.5 pA/pF ( p from four experiments such as in A. O, Open channel levels; C, closed channel levels. 0.05; n ϭ 14), and 6.5 Ϯ 1.0 pA/pF ( p Ͻ 0.001; n ϭ 19). We interpret the reduced currents in the PLC␦-PH cells as being sequence of the tyrosine kinase Lyn that achieves its plasma mem- attributable to the reduced abundance of free PIP2, resulting in a brane localization (Raucher et al., 2000). Because all of these reduced tonic maximal Po of the channels, consistent with the ␦ PIP2-testing constructs are linked to EGFP, we can use single-cell inability of PLC PH to bind to channels and stabilize their open- imaging to (1) confirm their presence in the cell and (2) localize ing. The Lyn-PH-PP-expressing cells are predicted to have a se- • ϩ 9832 • J. Neurosci., October 26, 2005 • 25(43):9825–9835 Li et al. PIP2 Regulation of Unitary KCNQ K Channels

verely depleted PIP2 abundance from the strongly augmented PIP2 phosphatase activity and the strongly reduced current den- sity to arise from a much lower Po of the channels. We verified that expression of the PIP2-binding constructs does not affect expression of channel protein using immunoblots of whole-cell lysates from cells expressed with myc-tagged Kv7.2ϩKv7.3 to- gether with each construct (data not shown). Assuming that the

effects on current amplitudes are as a result of reduction in the Po, we can use the dose–response curve of unitary Kv7.2/7.3 Po ver- sus [diC8-PIP2] (Fig. 3) to estimate changes in tonic membrane PIP2 caused by expression of the two PIP2 sequestering/degrad- ing constructs. Assuming a normal maximal Kv7.2/7.3 Po of 0.31 ␮ from our data, that corresponds to a [diC8-PIP2]of23 M in our inside-out patch experiments. In cells expressing PLC␦-PH, the

whole-cell currents were reduced by 69%, corresponding to a Po ␮ of 0.096 and a [diC8-PIP2]of8.2 M; in cells expressing Lyn-PH- PP, whole-cell currents were reduced by 91%, corresponding to a ␮ Po of 0.026 and a [diC8-PIP2]of3.0 M. Thus, under the assump- tion that changes in [diC8-PIP2] and in [PIP2] have similar ef- fects, we estimate that expression of PLC␦-PH and Lyn-PH-PP Figure 7. Effects of muscarinic stimulation on anionic phospholipids in CHO cells expressing reduced free PIP2 in CHO cells by 64 and 87%, respectively. M1 receptors.Anionicphospholipidsweredeterminedfromtheirdeacylationproductsbyanion exchangeHPLCwithsuppressedconductivitydetection.Dataarecompiledforapplicationofthe cholinergicagonistcarbachol(100␮M)for1or10min.BarsshowthemeanPIP ,PIP,PI,orPA, Stimulation of Gq/11-coupled M1 receptors depletes PIP2 in 2 CHO cells expressed as a percentage of total membrane phosphoinositides. The left axis refers to %PIP2 As discussed previously, muscarinic suppression of M-type cur- and %PIP, and the right axis refers to %PI and %PA.

rents is thought to primarily involve depletion of membrane PIP2 by activation of PLC (Suh and Hille, 2002; Zhang et al., 2003; Suh exposure, respectively. The %PI, however, continued to fall dur- et al., 2004). Our findings here suggest that the Po of these chan- ing maintained receptor stimulation, falling to 39.0 Ϯ 1.1% (n ϭ nels are highly sensitive to PIP2 abundance but that a strong 5) after 10 min of carbachol exposure. The %PA also continued to depletion of PIP2 would be required to cause the profound rise during the maintained stimulation, rising further to 24.1 Ϯ (Ͼ70%) suppression of the currents that is seen. Because this 1.4% (n ϭ 5) after 10 min of carbachol exposure. Both the con- hypothesis of how muscarinic stimulation suppresses M channels tinued fall in %PI and the continued rise in %PA suggest that

predicts a strong reduction in PIP2, we took advantage of a CHO PIP2 hydrolysis continued unabated during the longer carbachol cell line stably expressing M1 receptors to analytically measure exposure, arguing against receptor desensitization as the reason

changes in phosphoinositide abundance induced by muscarinic for the partial recovery of %PIP2 and %PIP during the 10 min stimulation of these cells. Using HPLC analysis, we assayed the treatment. Perhaps this partial recovery is attributable to the

abundance of several different phosphoinositides, including agonist-mediated stimulation of PIP and PIP2 synthesis by PI4- PIP2, phosphatidylinositol 4-phosphate (PIP), PI, and PA, before kinase and PI(4)5-kinase, as has been shown recently for Gq/11- and after exposure of the cells to the cholinergic agonist carbachol coupled bradykinin receptors in neuroblastoma cells (Xu et al., (100 ␮M) for 1 or 10 min (Fig. 7). 2003) and suggested in sympathetic neurons (Gamper et al., We found that muscarinic stimulation for 1 min indeed re- 2004). duced PIP2 abundance. Results are given as percentages of total anionic phospholipids, a pool that includes the phosphatidylser- Discussion ine and cardiolipin moieties that are not involved in PLC activity. The importance of phosphoinositides in cellular signals triggered Ϯ ϭ The %PIP2 was strongly reduced from 0.80 0.02% (n 7) by plasma-membrane receptors coupled to the Gq/11 family of before to 0.10 Ϯ 0.04% (n ϭ 5) after 1 min of carbachol exposure. G-proteins has been appreciated for decades. In this pathway, Receptor stimulation also caused a strong decline in %PIP, from hormonal stimulation results in activation of PLC, which hydro- Ϯ ϭ Ϯ ϭ 0.99 0.06% (n 7) before to 0.19 0.04% (n 5) after 1 min lyzes PIP2 into membrane-bound DAG and soluble of carbachol exposure. This parallel decline in %PIP2 and %PIP is trisphosphate, both potent intracellular signaling molecules. similar to that measured in human neuroblastoma cells (Willars However, in the past decade, it has become clear that the PIP2 et al., 1998). Unexpectedly, the %PI also displayed a significant molecule functions as a second messenger in its own right. Much decline, falling from 49.5 Ϯ 0.7% (n ϭ 7) before to 46.5 Ϯ 0.3% evidence has accumulated over the past several years suggesting ϭ (n 5) after 1 min of carbachol exposure. Because diacylglycerol that Kv7 channels are sensitive to the abundance of PIP2 in the (DAG), the membrane-bound byproduct of PIP2 hydrolysis, is plasma membrane (Delmas and Brown, 2005). Although Kv7 ϩ rapidly converted to PA by DAG-kinase, the %PA increased after channels are voltage-gated K channels (unlike PIP2-sensitive Ϯ receptor stimulation, as expected. The %PA rose from 15.8 Kir channels), the presence of PIP2 would act to stabilize opening 1.1% (n ϭ 7) before to 17.6 Ϯ 0.9% (n ϭ 5) after 1 min of by favoring the allosteric conformational change that opens the

carbachol exposure. Although muscarinic suppression of M/Kv7 channel, such that Po is increased at all voltages. The physiological channels does not noticeably desensitize in patch-clamp experi- role of such PIP2 sensitivity is perhaps most clear for the case of ments, we found the abundance of PIP2, and to a lesser extent M-type channels, which are thus regulated by depletions of mem- PIP, to slowly recover during maintained muscarinic stimula- brane PIP2 produced by PLC activation downstream of Gq/11- Ϯ tion. Thus, the %PIP2 and the %PIP recovered to 0.57 0.15% coupled receptor stimulation. This mechanism predicts the max- ϭ Ϯ ϭ (n 5) and 0.40 0.06% (n 5) after 10 min of carbachol imal Po of Kv7 channels in intact cells to be governed by the tonic • ϩ Li et al. PIP2 Regulation of Unitary KCNQ K Channels J. Neurosci., October 26, 2005 • 25(43):9825–9835 • 9833

abundance of PIP2 in the membrane and increases or decreases in channels, the large effects on their PIP2 regulation by neutraliza- tonic PIP2 levels to increase or decrease maximal channel Po, tions of positively charged amino acids lends support to the idea respectively. Stimulation of other Gq/11-coupled receptors in of direct binding, as does a recent crystal structure (Pegan et al., sympathetic neurons also suppress M currents, however not by 2005). Thus, direct binding to Kv7 channels is our working hy-

large changes in membrane PIP2 (Gamper et al., 2004) but by pothesis. For both families of channels that contain members 2ϩ generation of intracellular Ca signals (Cruzblanca et al., 1998; with divergent apparent affinities for PIP2, it seems as yet un- Bofill-Cardona et al., 2000), which may act in concert with cal- known whether their biochemical binding affinities are so diver- modulin (Gamper and Shapiro, 2003; Gamper et al., 2005b) to gent, or whether their intrinsic energy of opening diverges, result-

alter the affinity of the channels for PIP2, such that tonic PIP2 ing in a differential amount of stabilization energy coming from abundance is insufficient to keep most channels open (Winks et PIP2 binding being required to achieve the same level of channel al., 2005). opening. Because K ϩ channels are tetramers, and presumably

Here, we directly tested the PIP2-gating hypothesis by a variety each subunit can bind PIP2, it will be critical to learn the stoichi- of methods. The first was inside-out patch recording and a range ometry between PIP2 molecules and channel and the relationship of water-soluble diC8-PIP2 concentrations applied to the cyto- between bound PIP2 molecules/channel and opening. Work plasmic face of the patch. We found that the maximal single- from the Logothetis laboratory has also demonstrated direct gat-

channel Po of Kv7.2–7.4 and Kv7.2/7.3 channels is highly gov- ing of Kv7 channels by PIP2 in oocyte-excised macropatches erned by [diC8-PIP2], with a strongly differential apparent (Zhang et al., 2003). Their EC50 value for activation of Kv7.2/7.3 ␮ affinity of the channels for diC8-PIP2 that parallels the differen- channels by diC8-PIP2 was 87 M, with a Hill slope of 1.35, in the tial maximal Po in cell-attached patches (Li et al., 2004). In this range of that seen here for regulation of unitary channel Po. They study, we did not posit that the affinity of the channels for diC8- also showed direct regulation of Kv7.1, Kv7.4, and Kv7.5 chan-

PIP2 is the same as for (full-length) PIP2, but did assume that nels by PIP2, but they did not investigate any differential apparent relative differences in the apparent affinities of the different chan- affinities among the channels. The mild cooperativity for the

nels for diC8-PIP2 and full-length PIP2 do correspond. For all of binding of multiple PIP2 molecules to the channel modeled by the channels studied in inside-out patches, the Po could be in- Suh et al. (2004) seems in the correct range both for the slope of creased well beyond their values normally seen in cell-attached the dose–response curves and of muscarinic agonist versus M

patches by increasing the diC8-PIP2 concentration. Consistent current seen by many laboratories and for diC8-PIP2 versus ϭ with this result is the strong increase in maximal channel Po of single-channel Po seen here (nHill 1–1.8). Kv7.2/7.3 and Kv7.2 observed in cell-attached patches from cells The maximal on-cell Po for Kv7.2/7.3 heteromultimers is seen ϳ overexpressing PI(4)5-kinase, resulting presumably from greater here to be 0.3, an intermediate value between low-Po Kv7.2 and tonic PIP2 abundance in the membrane. The increase in macro- high-Po Kv7.3 (Li et al., 2004), and to be in the range of 0.13–0.3 scopic Kv7.2, but not Kv7.3, currents by overexpression of by the Brown laboratory (Selyanko et al., 2001; Tatulian and

PI(4)5-kinase was particularly dramatic and reinforces both the Brown, 2003). Our inside-patch data show that the maximal Po conclusions of low apparent affinity of Kv7.2 for PIP2 and the continues to rise past this range of values as the applied [diC8- highly differential maximal Po of Kv7.2 and Kv7.3 in cell-attached PIP2] is increased, and in on-cell patch, overexpression of PI(4)5- patches reported here and previously (Li et al., 2004). Conversely, kinase likewise strongly raises maximal Po. Together, these data maneuvers that tonically decreased the abundance of free mem- suggest that the tonic [PIP2] in intact cells is well under that brane PIP2 strongly lowered tonic current amplitudes, almost necessary to saturate PIP2 binding to the channels. However, in certainly by lowering channel Po. A strong reduction in unitary SCG neurons, the M channels of which heavily consist of Kv7.2/ channel Po by muscarinic stimulation was observed that corre- 7.3 heteromultimers, overexpression of PI(4)5-kinase did not in- lates well with the suppression of macroscopic currents, and that crease tonic M-current amplitudes (Winks et al., 2005), and we

resembles the changes in channel Po seen in the direct diC8-PIP2 found that its overexpression likewise did not significantly in- applications. Finally, HPLC analysis confirms the necessary pro- crease whole-cell Kv7.2/Kv7.3 currents expressed in CHO cells

found PIP2 depletion in cells heterologously expressing M1 re- (our unpublished observations). Perhaps this discrepancy be- ceptors. There is a discrepancy at later times, in that PIP2 and PIP tween single-channel and macroscopic effects is attributable to a levels recover in part over a few minutes during receptor stimu- secondary effect of increased [PIP2] in promoting of lation, whereas the inhibition of channel activity is usually stable membrane proteins, as has been described recently (Huang et al.,

over a similar time course. One explanation is that PIP2 is being 2004). Thus, an increase in current amplitudes by higher tonic generated in membrane domains from which PIP2 diffusion to M [PIP2] increasing channel Po might be balanced by a decrease in channels is in some way restricted. Despite this discrepancy, our tonic surface abundance of the channel proteins.

work is strongly supportive of PIP2 regulation of Kv7 channels as Left unanswered is the molecular mechanism by which PIP2 being physiologically integral to G-protein channel signaling. binding to the channel proteins stabilizes the allosteric confor-

Although our data suggest that the highly divergent Po values mational change that is opening. Because the “kink” in the sixth of Kv7 channels are associated with very differential apparent transmembrane segment (S6) thought to be the swivel point of

affinities for PIP2, we have no biochemical data on their binding the main gate of Kv channels (Jiang et al., 2002) is located in the affinities. Indeed, although several investigators assume that middle of the membrane, and because the S6 helix is thought to

PIP2-sensitive channels are regulated by direct binding of PIP2 end just after it emerges from the inner leaflet of the membrane, molecules to the channel proteins in the membrane, clear evi- PIP2 should bind to the channel downstream of S6 (i.e., to the dence for this remains sparse. However, if there is an intermedi- cytoplasmic tail). Indeed, mutagenesis (Zhang et al., 2003) and

ate protein that mediates PIP2 regulation of Kv7 channels, it single-channel recording of Kv7.3/Kv7.4 chimeras (Li et al., 2004; would have to be membrane associated because the regulation Gamper et al., 2005b) indicate that this is where PIP2-mediated happens in excised inside-out patches and must be present both regulation of channel opening occurs. From that point, the PIP2/ in the plasma membrane of mammalian cells (present study) and C-terminal interaction could result in some force that pulls the S6 of oocytes (Zhang et al., 2003). For the case of inward rectifier K ϩ helix away from the pore, opening it. Because Kv7 channels, like • ϩ 9834 • J. Neurosci., October 26, 2005 • 25(43):9825–9835 Li et al. PIP2 Regulation of Unitary KCNQ K Channels other Kv channels, are voltage gated, this must occur after the structure and mechanism of a -gated potassium channel. Nature movement of the voltage sensors of the channels permits gating. 417:515–522. Perhaps the PIP action is to anchor the post-S6 end of the chan- Kubisch C, Schroeder BC, Friedrich T, Lutjohann B, El-Amraoui A, Marlin S, 2 Petit C, Jentsch TJ (1999) KCNQ4, a novel potassium channel expressed in nel in place, causing the voltage-induced flexing of the channel to sensory outer hair cells, is mutated in dominant deafness. Cell 96:437–446. splay the pore open. Alternatively, the hydrophobic part of the Lerche C, Scherer CR, Seebohm G, Derst C, Wei AD, Busch AE, Steinmeyer K PIP2 molecule may cause it to favor being surrounded by lipid, (2000) Molecular cloning and functional expression of KCNQ5, a potas- generating a force away from the channel axis as the carboxy-tail sium channel subunit that may contribute to neuronal M-current diver- sity. J Biol Chem 275:22395–22400. is pulled from the rest of the protein. Such structural questions ϩ await more detailed knowledge of the molecular architecture of Li Y, Gamper N, Shapiro MS (2004) Single-channel analysis of KCNQ K Kv7 channels and of how PIP binding to the protein alters its channels reveals the mechanism of augmentation by a cysteine-modifying 2 reagent. J Neurosci 24:5079–5090. configuration in the plasma membrane. Maljevic S, Lerche C, Seebohm G, Alekov AK, Busch AE, Lerche H (2003) C-terminal interaction of KCNQ2 and KCNQ3 K ϩ channels. J Physiol (Lond) 548:353–360. References Maruyama T, Kanaji T, Nakade S, Kanno T, Mikoshiba K (1997) 2APB, Bender K, Wellner-Kienitz MC, Pott L (2002) Transfection of a 2-aminoethoxydiphenyl borate, a membrane-penetrable modulator of phosphatidyl-4-phosphate 5-kinase into rat atrial myocytes removes ϩ Ins(1,4,5)P3-induced Ca 2 release. J Biochem (Tokyo) 122:498–505. inhibition of GIRK current by endothelin and alpha-adrenergic agonists. Nasuhoglu C, Feng S, Mao J, Yamamoto M, Yin HL, Earnest S, Barylko B, FEBS Lett 529:356–360. Albanesi JP, Hilgemann DW (2002) Nonradioactive analysis of phos- Bofill-Cardona E, Vartian N, Nanoff C, Freissmuth M, Boehm S (2000) Two phatidylinositides and other anionic phospholipids by anion-exchange different signaling mechanisms involved in the excitation of rat sympa- high-performance liquid chromatography with suppressed conductivity thetic neurons by uridine nucleotides. Mol Pharmacol 57:1165–1172. detection. Anal Biochem 301:243–254. Cooper EC, Harrington E, Jan YN, Jan LY (2001) M channel KCNQ2 sub- Pegan S, Arrabit C, Zhou W, Kwiatkowski W, Collins A, Slesinger PA, Choe S units are localized to key sites for control of neuronal network oscillations (2005) Cytoplasmic domain structures of Kir2.1 and Kir3.1 show sites and synchronization in mouse brain. J Neurosci 21:9529–9540. for modulating gating and rectification. Nat Neurosci 8:279–287. Cruzblanca H, Koh DS, Hille B (1998) Bradykinin inhibits M current via ϩ ϩ Raucher D, Stauffer T, Chen W, Shen K, Guo S, York JD, Sheetz MP, Meyer T and Ca 2 release from IP -sensitive Ca 2 stores in rat 3 (2000) Phosphatidylinositol 4,5-bisphosphate functions as a second sympathetic neurons. Proc Natl Acad Sci USA 95:7151–7156. Delmas P, Brown DA (2005) Pathways modulating neural KCNQ/M (Kv7) messenger that regulates -plasma membrane adhesion. Cell potassium channels. Nat Rev Neurosci, in press. 100:221–228. Delmas P, Coste B, Gamper N, Shapiro MS (2005) Phosphoinositide lipid Roche JP, Westenbroek R, Sorom AJ, Hille B, Mackie K, Shapiro MS (2002) Antibodies and a cysteine-modifying reagent show correspondence of M second messengers: new paradigms for calcium channel modulation. ϩ 47:179–182. current in neurons to KCNQ2 and KCNQ3 K channels. Br J Pharmacol Etxeberria A, Santana-Castro I, Regalado MP, Aivar P, Villarroel A (2004) 137:1173–1186. Rohacs T, Chen J, Prestwich GD, Logothetis DE (1999) Distinct specificities Three mechanisms underlie KCNQ2/3 heteromeric potassium ϩ M-channel potentiation. J Neurosci 24:9146–9152. of inwardly rectifying K channels for phosphoinositides. J Biol Chem Ford CP, Stemkowski PL, Light PE, Smith PA (2003) Experiments to test the 274:36065–36072. role of phosphatidylinositol 4,5-bisphosphate in - Rohacs T, Lopes CM, Jin T, Ramdya PP, Molnar Z, Logothetis DE (2003) induced M-channel closure in bullfrog sympathetic neurons. J Neurosci Specificity of activation by phosphoinositides determines lipid regulation 23:4931–4941. of Kir channels. Proc Natl Acad Sci USA 100:745–750. Franke TF, Kaplan DR, Cantley LC, Toker A (1997) Direct regulation of the Sanguinetti MC, Curran ME, Zou A, Shen J, Spector PS, Atkinson DL, Keat- Akt proto-oncogene product by phosphatidylinositol-3,4-bisphosphate. ing MT (1996) Coassembly of KVLQT1 and minK (IsK) proteins to Science 275:665–668. form cardiac IKs potassium channel. Nature 384:80–83. 2ϩ Schwake M, Pusch M, Kharkovets T, Jentsch TJ (2000) Surface expression Gamper N, Shapiro MS (2003) Calmodulin mediates Ca -dependent ϩ modulation of M-type K ϩ channels. J Gen Physiol 122:17–31. and single channel properties of KCNQ2/KCNQ3, M-type K channels Gamper N, Stockand JD, Shapiro MS (2003) Subunit-specific modulation involved in epilepsy. J Biol Chem 275:13343–13348. Schwake M, Jentsch TJ, Friedrich T (2003) A carboxy-terminal domain de- of KCNQ potassium channels by Src tyrosine kinase. J Neurosci ϩ 23:84–95. termines the subunit specificity of KCNQ K channel assembly. EMBO Gamper N, Reznikov V, Yamada Y, Yang J, Shapiro MS (2004) Phosphati- Rep 4:76–81. Selyanko AA, Stansfeld CE, Brown DA (1992) Closure of potassium dylinositol 4,5-bisphosphate signals underlie receptor-specific Gq/11- mediated modulation of N-type Ca 2ϩ channels. J Neurosci M-channels by muscarinic acetylcholine-receptor stimulants requires a 24:10980–10992. diffusible messenger. Proc R Soc Lond B Biol Sci 250:119–125. Gamper N, Stockand JD, Shapiro MS (2005a) The use of Chinese hamster Selyanko AA, Hadley JK, Wood IC, Abogadie FC, Jentsch TJ, Brown DA ovary (CHO) cells in the study of ion channels. J Pharmacol Toxicol (2000) Inhibition of KCNQ1–4 potassium channels expressed in mam- Methods 51:177–185. malian cells via M1 muscarinic acetylcholine receptors. J Physiol (Lond) Gamper N, Li Y, Shapiro MS (2005b) Structural requirements for differen- 522:349–355. tial sensitivity of KCNQ K ϩ channels to modulation by Ca 2ϩ/calmodu- Selyanko AA, Hadley JK, Brown DA (2001) Properties of single M-type lin. Mol Biol Cell 16:3538–3551. KCNQ2/KCNQ3 potassium channels expressed in mammalian cells. Hadley JK, Passmore GM, Tatulian L, Al-Qatari M, Ye F, Wickenden AD, J Physiol (Lond) 534:15–24. Brown DA (2003) Stoichiometry of expressed KCNQ2/KCNQ3 potas- Shapiro MS, Roche JP, Kaftan EJ, Cruzblanca H, Mackie K, Hille B (2000) ϩ sium channels and subunit composition of native ganglionic M channels Reconstitution of muscarinic modulation of the KCNQ2/KCNQ3 K deduced from block by tetraethylammonium. J Neurosci 23:5012–5019. channels that underlie the neuronal M current. J Neurosci 20:1710–1721. Herlitze S, Garcia DE, Mackie K, Hille B, Scheuer T, Catterall WA (1996) Shaw G (1996) The pleckstrin homology domain: an intriguing multifunc- Modulation of Ca 2ϩ channels by G-protein beta gamma subunits. Nature tional protein module. BioEssays 18:35–46. 380:258–262. Stauffer TP, Ahn S, Meyer T (1998) Receptor-induced transient reduction Hille B (1994) Modulation of ion-channel function by G-protein-coupled in plasma membrane PtdIns(4,5)P2 concentration monitored in living receptors. Trends Neurosci 17:531–536. cells. Curr Biol 8:343–346. Huang S, Lifshitz L, Patki-Kamath V, Tuft R, Fogarty K, Czech MP (2004) Stolz LE, Kuo WJ, Longchamps J, Sekhon MK, York JD (1998) INP51, a Phosphatidylinositol-4,5-bisphosphate-rich plasma membrane patches yeast inositol polyphosphate 5-phosphatase required for phosphatidyl- organize active zones of endocytosis and ruffling in cultured adipocytes. inositol 4,5-bisphosphate homeostasis and whose absence confers a cold- Mol Cell Biol 24:9102–9123. resistant phenotype. J Biol Chem 273:11852–11861. Jiang Y, Lee A, Chen J, Cadene M, Chait BT, MacKinnon R (2002) Crystal Suh B, Hille B (2002) Recovery from muscarinic modulation of M current • ϩ Li et al. PIP2 Regulation of Unitary KCNQ K Channels J. Neurosci., October 26, 2005 • 25(43):9825–9835 • 9835

channels requires phosphatidylinositol 4,5-bisphosphate synthesis. Neu- Willars GB, Nahorski SR, Challiss RA (1998) Differential regulation of mus- ron 35:507–520. carinic acetylcholine receptor-sensitive polyphosphoinositide pools and Suh BC, Hille B (2005) Regulation of ion channels by phosphatidylinositol consequences for signaling in human neuroblastoma cells. J Biol Chem 4,5-bisphosphate. Curr Opin Neurobiol 15:370–378. 273:5037–5046. Suh BC, Horowitz LF, Hirdes W, Mackie K, Hille B (2004) Regulation of Winks JS, Hughes S, Filippov AK, Tatulian L, Abogadie FC, Brown DA, Marsh KCNQ2/KCNQ3 current by G-protein cycling: the kinetics of receptor- SJ (2005) Relationship between membrane phosphatidylinositol-4,5-

mediated signaling by Gq. J Gen Physiol 123:663–683. bisphosphate and receptor-mediated inhibition of native neuronal M Tatulian L, Brown DA (2003) Effect of the KCNQ potassium channel channels. J Neurosci 25:3400–3413. opener retigabine on single KCNQ2/3 channels expressed in CHO cells. Xu C, Watras J, Loew LM (2003) Kinetic analysis of receptor-activated J Physiol (Lond) 549:57–63. phosphoinositide turnover. J Cell Biol 161:779–791. Wang HS, Pan Z, Shi W, Brown BS, Wymore RS, Cohen IS, Dixon JE, McK- Zhang H, Craciun LC, Mirshahi T, Rohacs T, Lopes CM, Jin T, Logothetis DE

innon D (1998) KCNQ2 and KCNQ3 potassium channel subunits: mo- (2003) PIP2 Activates KCNQ channels, and its hydrolysis underlies lecular correlates of the M-channel. Science 282:1890–1893. receptor-mediated inhibition of M currents. Neuron 37:963–975.